[U-Boot] [U-BOOT] [PATCH] arm: fix flush cache function on arm926ejs

[Lei Wen]

On Wed, Jan 12, 2011 at 12:27 PM, Lei Wen <adrian.wenl at gmail.com> wrote:
> Hi Albert,
>
> On Wed, Jan 12, 2011 at 3:53 AM, Albert ARIBAUD <albert.aribaud at free.fr> wrote:
>> Hi,
>>
>> Le 11/01/2011 16:40, Lei Wen a écrit :
>>> flush_cache function should only be called when the dcache is on.
>>> And original flush method for arm926ejs seems don't work, replace
>>> it with new version.
>>>
>>> Test on Marvell Pantheon pxa920 board with usb ether function.
>>>
>>> Signed-off-by: Lei Wen<leiwen at marvell.com>
>>> ---
>>>   arch/arm/lib/cache.c |    8 ++++----
>>>   1 files changed, 4 insertions(+), 4 deletions(-)
>>>
>>> diff --git a/arch/arm/lib/cache.c b/arch/arm/lib/cache.c
>>> index 30686fe..b8d5eee 100644
>>> --- a/arch/arm/lib/cache.c
>>> +++ b/arch/arm/lib/cache.c
>>> @@ -27,16 +27,16 @@
>>>
>>>   void  flush_cache (unsigned long dummy1, unsigned long dummy2)
>>>   {
>>> +     if (!dcache_status())
>>> +             return;
>>>   #if defined(CONFIG_OMAP2420) || defined(CONFIG_ARM1136)
>>>       void arm1136_cache_flush(void);
>>>
>>>       arm1136_cache_flush();
>>>   #endif
>>>   #ifdef CONFIG_ARM926EJS
>>> -     /* test and clean, page 2-23 of arm926ejs manual */
>>> -     asm("0: mrc p15, 0, r15, c7, c10, 3\n\t" "bne 0b\n" : : : "memory");
>>> -     /* disable write buffer as well (page 2-22) */
>>> -     asm("mcr p15, 0, %0, c7, c10, 4" : : "r" (0));
>>> +     /* clean&  invalidate all D cache */
>>> +     asm("mcr p15, 0, %0, c7, c14, 0" : : "r" (0));
>>
>> ARM's ARM926EJ-S r0p4/r0p5 TRM does not define "... c7, c14, 0". Where
>> did you find this exact cp15 instruction?
>>
>
> My mistake...
> Pantheon board's sheeva core modify the standard cp15 op code a bit...
> With one opcode, it could invalidate and clean the dcache all.
>
> So could I modify the original patch like below?
> asm("mcr p15, 0, %0, c7, c6, 0" : : "r" (0));
> This would only invalidate all dcache. Since current dcache enable is
> just write-through,
> I think it is safe to just do the invalidate operation.
>
> This also could works for my board.

I take my words back...
I forget to open the dcache... After dcache is on, the usb tftp still
cannot works...
Seems still do the clean cache operation...

Best regards,
Lei